Structural imaging of the retina in-vivo

Get Complete Project Material File(s) Now! »

Retinal structure and function

Cellular origanization

As mentioned before, the retina detects the light signal and converts it to nerve impulses. This organ is therefore composed of photosensitive cells connected to neuronal cells. Fig-ure 1.5 shows an illustration of the cellular organization of the retina.The region con-taining synapses linking the photoreceptors with bipolar and horizontal cell dendrites is known as the outer plexiform layer or Outer Nuclear Layer (ONL), the area where the bipolar and amacrine cells connect to the ganglion cells is the inner plexiform layer or In-ner Nuclear Layer (INL). The inner nuclear layer, contains one to four types of horizontal cells, 11 types of bipolar cells and 22 to 30 types of amacrine cells. The surface layers of the retina are the Ganglion Cells Layer (GCL), containing about 20 types of ganglion cells and the nerve fiber layer (NFL), going from the eye to the brain through the optic nerve head.
These four neural layers are separated by three vascular layers, the Superficial, In-termediate and Deep layers. Under the ONL lies the photoreceptor layer identified in Fig.1.5(B) as Inner Segments (IS)/ Outer Segments (OS). Finally, the last retinal layer over the choroid is the Retinal Pigment Epithelium (RPE).
FI G U R E 1.5: (A) Schematic drawing showing a cross-section of an eyeball in a top view. (B) Schematic layout showing a cross-sectional view of the retinal and choroidal layers.[43]
The following paragraphs provide more details about the retinal structures, between the ones described above, that are addressed in this work:
The axons of the optic nerve are extensions of the retinal ganglion cells whose unmyeli-nated axons form much of the nerve fiber layer of the neural retina. The axons or “nerve fibers” then enter the optic disc by making a sharp turn, where they continue as a series of fascicles or bundles [44]. Nerve fiber bundles from a group of retinal ganglion cells travel together with little tendency to disperse laterally as they course into the optic nerve head [45], which collects all the axons of the ganglion cells.


The photoreceptors are light sensitive cells and there are two categories: cones and rods. Cones, responsible for daylight vision have a higher density in the central region of the retina, the macula, to ensure accurate vision. The fovea, the center of the macula, contains only cones. The cone photopigments (opsins) are sensitive to a range of wavelengths. There are three type of cones, Long, Middle and Short wavelength sensitive, depending on their peak wavelength. Cones diameter is 1-4μm at the fovea and over 4μm outside this region. Rods, more adapted to night vision, have approximately a diameter of 1μm. Their density is maximum at approximately 6 mm from the center of the retina and decreases towards the periphery of the retina (see Fig.1.6 (B))[46].
Photoreceptors are composed of three main layers: 1) the cell nuclei, 2) the inner seg-ment, which encloses the cell machinery, and 3) the outer segment that contains photo-sensitive pigment (see Fig. 1.6 (A)). The outer segment corresponds to discrete disks satu-rated with a photosensitive molecule called rhodopsin in the case of the rods, and forms a series of folds containing similar photosensitive molecules in the case of cones. The outer segment absorbs photons, which initiates an electrochemical transmission through the cells and retinal nerve fibers, up into the brain [47].
FI G U R E 1.6: (A) Schematic illustration of a cone (left) and a rod (right)
showing the three main sections: 1) the outer nuclear layer,containing the nuclei 2) the inner segment and 3) the outer segment that contains photo-sensitive pigment.[47] (B) Density of rod and cone photoreceptors across the human retina as a function of eccentricity (in degrees of visual angle relative to the position of the fovea) for the left eye. Cones are concen-trated in the fovea while rods are absent from the foveal region and reach their highest density 10 to 20 degrees peripheral to the fovea. No photore-ceptors are present in the blindspot.[46]

Retinal pigment epithelium (RPE)

The RPE is the pigmented cell layer located between the light-sensitive outer segments of the photoreceptors and blood supply of the choroid. The RPE is composed of a single layer of hexagonally packed, tight-junction, connected, single sheet of cells with pigment granules and organelles for digestion of photoreceptor outer segment membranes into phagosomes [48].

Vascular network

The central retinal artery and the choroidal blood vessels are the two sources of blood sup-plying the human retina. The largest blood flow is provided by the choroid (65-85%) and maintain principally the outer retina, and photoreceptors in particular. The retinal artery provides the rest of the blood flow (20-30%) and enters the retina through the optic nerve head, supplying the inner layers[49]. Figure 1.7 shows the vascular network model of the retina proposed by Campbell et al.[50](left) and Fouquet et al.[51](right). In both studies the central retinal artery supplies four retinal vascular networks: 1) The radial peripapil-lary capillaries (RPCs), which is the most superficial layer of capillaries lying in the inner part of the nerve fiber layer (this layer is presented in [50] but does not appear in Fig. 1.7(A) as the images are not acquired near the optic disk, where the RPC layer lies). 2) The Super-ficial Vascular Plexus, located at the ganglion cell layer. 3) The Deep Capillary Plexus or Intermediate Vascular Plexus located above the inner nuclear layer. 3) The Deep Capillary Plexus or Deep Vascular Plexus, located below the inner nuclear layer.
FI G U R E 1.7: (A) Middle bottom panel: Color fundus photograph. Left panels: En-face PR-OCTA, which are arrayed from the most superficial on top to the deepest at the bottom. Right panels: En-face PR-OCTA slabs. Middle top panel: Cross-sectional PR-OCT. Center panel: Cartoon depict-ing the anatomical relationships between arterial and venous systems in the three vascular plexuses and the interconnecting layers.[50]. (B) Pro-posed 2D model of the retinal microcirculation in the pig retina close to the optic disc [51]. RPCs: Radial peripapillary capillaries, SVP: Superficial Vascular Plexus, IVP: Intermediate Vascular Plexus, DVP: Deep Vascular Plexus.

Retinal structure and function 

The visual cycle

The processing of visual information begins in the retina with the detection of light by photoreceptor cells. In humans, two specialized types of photoreceptors detect light un-der different conditions. Rod photoreceptors are highly sensitive and mediate vision in dim light, while cone photoreceptors function in bright light and mediate both high acu-ity and color vision. To detect light, both rods and cones exploit the unique properties of a vital molecule, retinal, a photosensitive derivative of vitamin A which initiates the phototransduction and is produced by the retinal pigment epithelium (RPE) [52]. Retinal molecules are recycled back into the RPE.
FI G U R E 1.8: Schematic illustration of intricately wired neurons in the retina. Signals pass from the photoreceptors through a series of neu-ral connections toward the surface of the retina, where the ganglion-cell nerve-fiber layer relays the processed information to the optic nerve and into the brain.
Then, neurons in the outer layer respond to stimulation originated in the photorecep-tors and continues the transmission through the visual pathway to the inner layer and finally the ganglion cells. Finally, impulses from the ganglion cells travel to the brain via more than a million optic nerve fibers (cf.Fig. 1.8). Although the visual function is com-posed by the parallel or successive mechanisms described above, there are other struc-tures carrying functions just as critical to the balance of the visual cycle. For instance, the RPE the RPE has another role essential to the normal function and survival of photorecep-tors which is to facilitate photoreceptor cell renewal. Indeed, rod and cone photoreceptor cells undergo a daily regeneration process wherein about 10% of their OS volume is shed, subsequently phagocytosed by adjacent RPE cells [53]. Another key function to maintain this balance is the vascular perfusion which provides the retinal cells involved in the cycle in oxygen and nutrients.
All these sub-functions, either with direct or indirect intervention in the visual cycle, are involved in some way.
This thesis manuscript describes, among other developments, the first steps of func-tional imaging undertaken with the implementations carried out during these three years. From the multiple functions leading to the visual cycle or indirectly maintaining its phys-iological balance, the following are going to be addressed in this work:

Photoreceptor orientation

The photoreceptors carry out the transduction, which is the conversion of the light to electrical signal. They have therefore a key function in the visual cycle. In the case of a healthy retina they should be oriented towards the center of the pupil (with a deviation of less than a millimeter). It has been hypothesized that a loss of this orientation is associated with a loss in visual capacity. Even though this hypothesis has not been validated yet, we aim to study this orientation which we believe could be a biomarker to evaluate the correct functioning of the photoreceptor cells.


Perfusion, i.e. the supplying of oxygen and nutriments to retinal cells, is carried out by the vascular network. This structure does not directly participate in the visual cycle, but a failing in its function would affect structures which do. Therefore, in this work we aim to characterize this indirect function, in particular at the microvascular level where tissue exchanges take place.


Clinical retinal imaging systems

Fundus camera

A fundus camera is a specialized low power microscope with an attached camera. The main structures that can be visualized on a fundus photography are the central and pe-ripheral retina, optic disc and macula. Fundus photography can be performed with col-ored filters, or with specialized dyes including fluorescein and indocyanine green. Its op-tical design is based on the indirect ophthalmoscope. Fundus cameras are described by the angle of view – the optical angle of acceptance of the lens. An angle of 30°, considered the normal angle of view, creates a film image 2.5 times larger than life.


Optical Coherence Tomography (OCT), consists in producing interferences between the light coming back from the sample (the retina in our case) and a part of the illumina-tion beam used as a reference. By modulating the wavelength (Fourier or Spectral Do-main OCT) or the position of the reference beam (Time Domain OCT), the recorded in-terference pattern gives access to propagation time difference between the sample and the reference beams, and thereby on the depth of the layers constituting the retina. This technique enables an axial resolution worth a few microns, typically ten times better than classical (incoherent) imaging.

High resolution retinal imaging

Adaptive Optics

Adaptive optics enables the real time correction of the wavefront aberrations introduced by the optics of the eye thanks to the following components:
1) The Wavefront sensor (WFS) which measures the wavefront aberrations. From the range of existing WFS, the most commonly used is the Hartmann-Shack, which is the one that will be used in this work.
2) The Wavefront corrector, in our case a deformable mirror, which compensates the measured aberrations introduced by the optics of the eye and the optical system.
3) The Real time calculator (RTC) which converts the measures into voltage commands and is used to pilot the wavefront corrector.


The principle of adaptive optics applied to retinal imaging is presented in Fig.2.1. The sys-tem is composed of the elements of the AO loop previously mentioned forming a closed loop configuration with the deformable mirror upstream from the WFS, a beacon illumi-nating the retina and an imaging camera. The illumination beacon is known as the WFS source and creates a point source in the retina. This illumination beacon is then back-scattered by the retina through the eye and propagates through the optical system towards the WFS. The deformation measured by the WFS is transmitted to the real time calculator which computes the voltage commands that are sent to the deformable mirror. Both the WFS and deformable mirror are placed at the pupil plane of the system. The deformable mirror surface is modified in order to compensate the mesured aberrations of the WFS. This correction is applied to the wavefront of the back-scattered photons coming from the large illumination source leading to a high resolution image of the retina.
Adaptive Optics was translated to retinal imaging in 1997 by Liang et al. [7]. They used adaptive optics (AO) to compensate for monochromatic aberrations of the eye. The quality of retinal images obtained on a CCD fundus camera was greatly improved so as to visualize cellular structures of just a few microns such as photoreceptors in human retina in-vivo. AO-FIO devices were further improved and their applications multiplied leading to a commercial system developed by the French company Imagine Eyes used now in a clinical setting [54].

High resolution retinal imaging 

The adaptive technology was subsequently introduced in scanning laser ophthalmo-scope (SLO) as well. The SLO was invented by Webb et al.[55] and represented a major ad-vance with respect to the wide field ophthalmoscope. This system presents an improved efficiency in light collection and real-time imaging, in particular thanks to the use of a con-focal pinhole which enhances the image contrast compared to flood illuminated imaging as well as optically sectioning the retina. By adding AO technology to the SLO [8], the SLO’s spatial resolution, of 20 μm and 200 μm lateral and axial respectively , evolved to 2.5 μm and 100 μm.
The introduction of a confocal pinhole in microscopy led to an improvement in image quality arising from an enhancement in spatial resolution as well as optical sectioning[56]. The ophthalmoscope combined with the lens of the eye can be seen as a microscope imag-ing the retina and thus in 1987 Webb et al. developed the confocal scaning ophthalmo-scope improving the contrast of retinal images [57]. This confocal configuration was later implemented on the AO-SLO leading to an instrument acquiring highly contrasted images of microstructure of the retina in-vivo. However, as mentioned before, these confocal sys-tems are limited by the speed of the resonant scanner, which additionally is slower than the involuntary fixational movements of the eye leading to distortion artifacts.
Another confocal regime, known as line scanning ophthalmoscopy, has been devel-oped to try to overcome these limitations as well as to simplify the complexity of these scanning systems.
Several variations of line scanning imaging systems have been investigated to improve the imaging speed and applied first to the cSLO and then to the cAOSLO in order to ren-der these systems less susceptible to eye motion. The idea first introduced by Hammer et al.[58, 59] was to use a line illumination instead of a « flying spot » and named this new configuration the line scanning ophthalmoscope (LSO). This implementation simplified the system to a configuration where only one of the two scanners, a galvanometer, is used thus increasing the imaging speed with limited reduction in the confocality. Further sim-plification and speed improvement of the system was achieved by completely avoiding the use of scanners. The galvanometer is replaced by a Digital Micromirror Device (DMD), a spatial light modulator which consists in an array of micromirrors that can be set to two positions allowing the projection of specific illumination patterns. In the case of the LSO it is set to rapidly project a single line to the retina and synchronizing the acquisition with the rolling shutter of the camera [60, 61]. Finally, faster imaging was achieved by acquiring in parallel multiple lines within the field of view [29, 62].

Table of contents :

1 The Human retina 
1.1 The human eye
1.1.1 Optics of the eye
1.1.2 Eyemovements
1.1.3 Ocular aberrations
1.2 Retinal structure and function
1.2.1 Cellular origanization
1.2.2 The visual cycle
2 Structural imaging of the retina in-vivo 
2.1 Clinical retinal imaging systems
2.1.1 Fundus camera
2.1.2 OCT
2.2 High resolution retinal imaging
2.2.1 Adaptive Optics
2.2.2 Off-axis techniques
3 Functional imaging of the retina in-vivo
3.1 Definition of functional imaging
3.2 The ideal functional imaging technique for the retina
3.3 Scan versus Full-Field
II Adaptive optics system: optimization & add-on 
4 Optimization and exploitation of an AO FIO
4.1 Introduction to the peer-reviewed article
4.2 Peer-reviewed Article: High loop rate adaptive optics flood illumination ophthalmoscope with structured illumination capability
4.3 Conclusion
5 Designing a new illumination for AO-FIO
5.1 Inspiration frommicroscopy
5.1.1 Oblique back-illumination
5.1.2 Structured IlluminationMicroscopy
5.2 Building a new illumination path
5.2.1 Deriving two illumination geometries
5.2.2 Optical design
5.2.3 Implementation
5.2.4 Difficulties and limitations
5.3 Structured Illumination Imaging
5.4 Conclusion
III Imaging instruments: Design, integration & testing 
6 Developing a dark-field modality in full field ophthalmoscope 
6.1 Introduction to the article
6.2 Article in Progress: Dark-field imaging using an Adaptive Optics Flood Illumination Ophthalmoscope
6.3 Conclusion
7 Developing pseudo-confocal modality 
7.1 Description of pseudo-confocal imaging technique
7.1.1 Principle of confocality and near-confocality
7.1.2 Implementation in the PARIS AO-FIO
7.2 Proof of concept in the photoreceptor layer
7.2.1 Optical sectioning
7.2.2 Limitations
7.3 Conclusion
8 Developing a Retinal Goniometer
8.1 Description of the technique
8.1.1 Phase imaging in the retina by oblique illumination
8.1.2 Implementation in the PARIS AO-FIO
8.2 Proof of concept in artificial eye
8.3 Test in humans
8.4 Conclusion
IV Clinical application: From structure to function 
9 Extracting neural retinal biomarkers
9.1 Structural biomarkers
9.1.1 Photoreceptors
9.1.2 Nerve fibers
9.1.3 Comparison with AO-SLO
9.2 Functional biomarkers
9.2.1 Clinical interest
9.2.2 Biomarker: Photoreceptor brightness at various illumination incidences
9.3 Conclusion
10 Extracting vascular retinal biomarkers
10.1 Structural biomarkers
10.1.1 Clinical interest
10.2 Functional biomarkers
10.2.1 Reading guide
10.2.2 Peer reviewed Article: Near Infrared Adaptive Optics Flood Illumination Angiography
10.3 Conclusion
11 Pigment epithelium imaging with two complementary techniques 
11.1 Clinical application of the AO-FIO dark-field images
11.1.1 Clinical interest
11.1.2 Retinal pigment epithelium biomarkers
11.1.3 Extraction of density biomarker of the RPE mosaic
11.1.4 Comparison with AO SLO
11.1.5 Conclusion
11.2 Autofluorescence imaging of the retinal pigment epithelium with AO-SLO
11.2.1 Reading guide
11.2.2 Peer-reviewed Article: In vivo near-infrared autofluorescence imaging of retinal pigment epithelial cells with 757nm excitation
11.3 Comparing two complementary imaging techniques of RPE
11.4 Conclusion: Towards functional imaging


Related Posts